BestMate MCP Server

This MCP (Model Context Protocol) server integrates BestMate's prompt optimization capabilities directly into your IDE through Cursor.

Quick Install (Recommended)

npm install -g bestmate-mcp-server

Setup

1. Generate your API key:

   • Go to https://www.bestmate.io/
   • Navigate to the "API Keys" tab
   • Click "Generate API Key"
   • Copy the generated key

2. Configure Cursor IDE: Add to your Cursor MCP settings:

   {
     "mcpServers": {
       "bestmate": {
         "command": "bestmate-mcp",
         "env": {
           "BESTMATE_API_KEY": "your-api-key-here"
         }
       }
     }
   }

Alternative: Development Install

If you want to install from source:

git clone [your-repo-url]
cd promptloop/mcp/bestmate
npm install
npm run build

# Configure Cursor with local path:
{
  "mcpServers": {
    "bestmate": {
      "command": "node",
      "args": ["./path/to/promptloop/mcp/bestmate/dist/index.js"],
      "env": {
        "BESTMATE_API_KEY": "your-api-key"
      }
    }
  }
}

Available Tools

bestmate_submit_prompt

Submit a prompt to BestMate for optimization analysis.

• Input: prompt (string, required)
• Optional: context, domain, model, temperature, optimization_type (all preset to sensible defaults; you do not need to provide these unless you want to override)
• Output: sessionId for tracking the optimization

Tip: In Cursor, you only need to enter your prompt. All other settings are optimized for you by default.

bestmate_get_results

Retrieve optimization results for a session.

• Input: sessionId (string)
• Output: Optimization suggestions and improvements

bestmate_apply_optimization

Apply a selected optimization to your prompt.

• Input: suggestionId (string), optimizedPrompt (string)
• Output: Confirmation of applied optimization

bestmate_evaluate_prompt

Evaluate a prompt for clarity, effectiveness, and specificity.

• Input: prompt (string), context (optional), criteria (optional array)
• Output: Evaluation scores and feedback

Workflow

1. Select prompt text in Cursor
2. Use bestmate_submit_prompt to start optimization (just enter your prompt!)
3. Use bestmate_get_results to get suggestions
4. Use bestmate_apply_optimization to apply chosen improvement
5. Use bestmate_evaluate_prompt to get performance scores

This replicates the exact BestMate browser workflow but keeps you in your IDE!